Screen to Text to Speech
🔊 sttts — Turn Any Screen Into an Audiobook (100% Local AI, No
Cloud)
Watch sttts read a Kindle book hands-free — OCR reads the page,
Kokoro speaks it aloud, then automatically clicks "next page" and
repeats. No APIs. No subscriptions. Runs entirely on your own
machine.
🧠 HOW IT WORKS
1. Draw a rectangle over any screen region
2. LightOnOCR-2-1B reads the text (AMD GPU / ROCm)
3. Kokoro-82M speaks it through your speakers (CPU)
4. Pixel diff skips frames where nothing changed
5. Auto-clicks "next page" when done reading
⚡ FEATURES
✅ 100% local — no API keys, no cloud
✅ Works on Kindle, PDFs, epub readers, web pages
✅ Smart idle detection — OCR only fires on changes
✅ Hands-free auto page-turn
✅ AMD GPU support via ROCm (RX 7900 XTX)
🔗 LINKS
GitHub: github.com/paradisecy/sttts
OCR Model: huggingface.co/lightonai/LightOnOCR-2-1B
TTS Model: huggingface.co/hexgrad/Kokoro-82M
DEV.to: dev.to/paradisecy/i-built-a-local-screen-reader-that-reads-your-screen-aloud-no-cloud-no-api-keys-m9
#LocalAI #TextToSpeech #OCR #Python #OpenSource #Kokoro #AMD #ROCm
#ScreenReader #Accessibility #Kindle #AI